Netflix and Google Make Land Grab On Edge of Internet 85
An anonymous reader writes "In an end-run around slow Internet backbone providers, Netflix and Google (plus a dozen more large content giants) are in a bitter fight to deploy servers and dominate the consumer edge of the Internet. This Wired article provides some of the first graphics of this fight and how it is changing the underlying Internet infrastructure. The source of the article (DeepField blog post) also has some pretty interesting commentary."
basically vertical integration with CDNs (Score:5, Informative)
As the article notes, from an internet-topology standpoint this isn't that new, dating back to Akamai-type CDNs starting in the 1990s. The idea is that you mirror your content inside several of the major edge networks, so e.g. Comcast users get served from the Comcast-local mirror. You then update the mirror whenever there's new content, but every single user doesn't have to re-fetch that video over the public internet to Comcast's network.
The main difference is that some of the large content providers are building out their own private CDNs, so Google is setting up its own edge-network mirrors instead of contracting out to Akamai. That's not a major technical change, but could have some important implications for competition.
Re:basically vertical integration with CDNs (Score:5, Insightful)
There could be some implications, but I don't think it's that bad. Akamai isn't going anywhere, they actually never delivered Netflix, and have had only limited business with Google, so it's not like they're "losing" anything. So if a competitor to any Google or Netflix comes along, they'll have the option of using Akamai until they're big enough for their own CDN, if that's actually the direction things are headed. On top of that, I would not be surprised if the day comes when Google starts offering its CDN as a service, which will actually add competition for Akamai.
Re:basically vertical integration with CDNs (Score:5, Interesting)
Yeah, I agree with that. Google and Netflix are basically moving a commodity service in-house, which often makes sense once you're using enough of it. But others can continue to use the commodity service, as long as it continues to exist, which it looks like it will.
If anything, the basic way bandwidth is billed and peering agreements are arranged is a bigger problem for small players than the edge caches this story is trying to get us worried about. The first problem someone is going to have if they try to compete with YouTube is not CDN access, but the fact that they would have to pay for bandwidth, whereas in many cases Google doesn't. For example, I work at a university, and our university network has a peering agreement with Google: that means that all YouTube watching by students is free to Google. A YouTube-competitor startup would have to pay for transit.
Re: (Score:2)
Re:basically vertical integration with CDNs (Score:4, Insightful)
Your use of the word 'free' is misplaced. This is vertical integration, plain and simple. In the early 1900s, Andrew Carnegie controlled the iron ore, steel manufacturing (made of iron), railroad tracks (made of steel), and railroad cars (to ship iron to manufacturing/selling, and steel for selling).
You wouldn't say that the iron was 'free' to Carnegie Steel because they owned the mines, would you? Content provisioning won't be 'free' to Google, as they have to buy/maintain the servers.
Re: (Score:2)
Content provisioning won't be 'free' to Google, as they have to buy/maintain the servers.
buy/maintaining the servers is "free" (or rather, as good as free) if the result of doing so saves sufficiently more money than the price of purchase+maintenance.
Re: (Score:1)
Actually, yes they do. From the article, netflixed used 3 CDN including akamai which evenly had traffic at about 33%. After it's deployment, other CDNs usage dropped to less then 10% for each while netflixed cdn accounted for 77.6%.
Will this effect akamai? Maybe but probably not too big, from the opinion of the blog post.
"Most CDNs would rather focus on higher value (and more profitable) services like analytics, acceleration, and DRM (e.g. see recent CDN product announcements)" from the blog post.
Re:basically vertical integration with CDNs (Score:5, Informative)
Re: (Score:2)
Agreed, not a major change. However, it might indeed affect competition. I'm glad AT&T has found a way to make money by helping large companies improve bandwidth rather than imposing a tiered system and effectively reducing service delivery for those who couldn't afford it. This attempt at a tiered internet sparked the net neutrality "movement" (if it qualifies as a movement, as such). The idea of private CDN's certainly is a more "positive" approach, but I do wonder if it introduces a precedent tha
Re:basically vertical integration with CDNs (Score:4, Informative)
I think that's the biggest reason for this change: They need to stay competitive with companies that don't have to pay for bandwidth. See Netflix is a bandwidth hog. Who will pay? (Hint: You.) [cnn.com]
Re: (Score:1)
WRONG, you are already paying for all the bandwidth you are using through your $20 - $50 monthly flat rate subscription, content providers should not have to pay anything on top of that some people even suggest that they should be paid by your ISP, the problem is your ISP even with earning millions, or even billions dollars wants more money and is trying to find ways to get bigger part of cookie (check what comcast is doing)
Re: (Score:2)
What they "should" have to pay is irrelevant; the reality is that (right now) they do have to pay for bandwidth unless they make deals like this.
Re: (Score:3)
Re: (Score:3)
The case for net neutrality (Score:5, Insightful)
That's not a major technical change, but could have some important implications for competition.
Yes, specifically that it'll fragment the entire network and potentially destroy interoperability across it. The internet would no longer be a unified global network. Network neutrality is the key to preventing this, but as we've seen, corporations don't want that: They want to turn the internet into a largely read-only media... just a better version of television.
A classic example of how this is shaping up is with Comcast, the Great Evil of the USA internet: They recently instituted a 250GB transfer limit, and then exempted Hulu from it, which they bought out. Netflix, a competing service at a lower price is now sitting out in the cold. Let's run some numbers and see how much of a problem this is. The average person watches 2.7 hours of TV per day; and it remains the single largest leisure activity in the United States. The average Netflix stream (based on my experience), is about 350KB/s. So that comes out to about 3.24GB per person, per day -- or 98.82 per month (the average length of a month). Now the number of people per household is a bit shaky, since there aren't any current numbers, but it's around 2.6 people per. So the average household will consume 257 GB per month if they used Netflix.
How strange that the bandwith cap is almost exactly the same number eh? Make no mistake -- this is a war between big business, and the only losers will be you and me. This is what happens when you let people into public positions who entertain the notion that capitalism runs best when it isn't regulated. Every infrastructure service in this country runs better with regulation, and the internet (telecommunications) is not an exception. Every time we let the private sector take over, we get crap like Standard Oil, AT&T (pre-breakup), Microsoft, etc. And now we have Montsano eating up our food supply (literally).
If network neutrality isn't given the force of law in the next two years, then two things are going to happen: Either we start building tunneled networks so all traffic through the last mile ISPs is encrypted and cannot be shaped, modified, or tampered with except in terms of bandwidth and latency as a whole... or we abandon the internet and start a new network that has no last mile restrictions (read: wireless, read: pirate radio).
Re: (Score:1, Insightful)
Slashdot is about 50% geeks these days, the other 50% are whiney technophopbic whiney pants as you can see evidenced here.
Want better network topology? WHAAAAA!!!
Sorry technophobic baby-man! We'll make the networks suck for everyone equally so you feel better.
Re: (Score:2)
I am troubled by the implications of capping, but let's keep the discussion grounded in reality.
I miss Ma Bell (Score:5, Interesting)
Re: (Score:2)
If you look at the upper right corner of the building in TFA's picture, you can see the shadow of the old Bell System logo.
I don't mis Ma Bell at all, because I can remember paying in the neighborhood of a dollar per minute for interstate toll calls, and that was in 1970's dollars. The telecom industry is reconsolidating and has what is arguably the most powerful lobby in Washington. Any way you look at it, that's bad. This twist is just as bad for smaller "web sites". Once the big players have moved all their stuff to their own servers near the edge, Akamai et all will be hurting. That's bad too.
Re: (Score:1)
The telecom industry is reconsolidating and has what is arguably the most powerful lobby in Washington
OK, I'll argue. I don't think the telecom industry's lobbying holds a candle to either the banking industry or the insurance industry.
Re: (Score:1)
How much of your life is data? How much of that data goes through the network of your bank and insurance company? Which banks and insurance companies siphoned your data and supplied it to the NSA in violation of US law? Which companies had to be "retroactively" absolved of any legal liability for breaking that law?
No. Telecom has much more influential lobbying since telecom is the bitch the government want's to have in it's bed. Information is power. With the correct access to information, I could put
Re: (Score:2)
Not that the banksters don't swing a big dick, but it's influence comes from a rather different vector than conventional "lobbying".
"changing the underlying Internet infrastructure" (Score:5, Interesting)
Re: (Score:2)
Re: (Score:2)
instead of having Akamai, or level3 record it for them? I'm confused what are on about? Seems like a non story to me. Big high bandwidth company moves some of it's bandwidth to the edges.
I'd bet that each netflix end doesn't have the full catalog, but just the most popular 500 things or something like that. They may even be doing something like that 1-2 second "start loading the movie but call it instant" on the most popular 10-15 is because they can get 5-10 requests for it in the same second and start all
This metric saddens me (Score:5, Insightful)
From the blog:
Our most recent data finds that more than 70% of all Internet traffic (on average) comes from just 150 CDN, hosting, cloud and content companies.
For many Internet users it is essentially now Television Mark 2.
Re: (Score:2)
Search Listing Problem (Score:2)
That's largely because you cannot acheive a search listing as a leaf node. If you don't point outward from you site you are judged to have no value.
This becomes a black-body radiation problem, New conent can go onto the radiant surface anywhere, but it gets lost in the geography becasue if that new content isn't part of a site containing a massive number of outbound links (usually ads and the "natrual link farm" of, say, youtube video pages having links to other youtube video pages) then your site is "benea
Re: (Score:2)
Re: (Score:3)
Re:Why don't they...? (Score:4, Interesting)
The actual protocol is agnostic - the client can ask for whatever chunk he wants. Most clients follow a (seemingly) random pattern, but nowadays uTorrent has a button you can click to prioritize the first blocks, so that you can start watching the movie while it's still downloading.
Re: (Score:2)
A client could request chunks in order. The issue becomes the latter parts of the file being less available than the earlier parts, especially if people don't watch/download the whole thing, or if they don't seed for very long.
I can't blame people for not seeding much--at least, many people. My ISP (AT&T DSL) has decreased service by adding bandwidth caps and raised rates, to boot. I've never had faster than 5 Mbps/768 Kbps Internet access (poor me), but some people have 30+ Mbps access with multi-me
Re: (Score:3)
While here the service has actually improved over the years (which, as you say, is what you'd expect), I got a VPS in the Netherlands with 300GB/month and 1Gbps up/down (burst) for less than $3/month and now I do all my seeding from there (as well as using it for my email server, hosting some very low traffic websites, etc).
Re: (Score:2)
Wow, that's twice my home bandwidth cap. Would you share the URL for them please?
Re: (Score:2)
OK, with the caveats that I'm not affiliated and I've owned the VPS for less than a month, so it's not my fault if they fuck up ;)
Oh, and it only has 7GB of disk. It's enough for me since I delete the stuff constantly (and with that upload speed, I always get a ratio greater than 3), but it's limited.
You can get 10GB more, but it's $3.7 per month - more than a new VPS!
So yeah, if you can live with that, they seem fine. If not, check http://www.lowendbox.com/ [lowendbox.com] that's where I found them.
Finally, Netherlands fo
Re: (Score:2)
Why would BT be harder to cache for a Bittorrent-aware cache?
And obviously you wouldn't have a pure BT solution that had to get its data from "the other side of the world", you'd have an hybrid solution with a location filter on the BT client and an HTTP fallback.
Re: (Score:2)
Net neutrality (Score:3)
Does this provide any leverage to help insure net neutrality by taking some power away from the core backbones, or is it moot since ISPs still reign supreme by providing that last mile connectivity?
Re: (Score:3)
Re: (Score:2)
Translation (Score:2)
We need to have bandwith pricing based on pipe size and not bits delivered per time period.
Of course since that would make every link symmetric there would be no chash-flow pressure.
Which is why internet connectivity should be a public good monopoly and not a competetive business.
The model itself, which was created once people started "charging for site access" and then the bandwith providers decided they wanted a piece of that action, is unsustainable because the Dance of the Sugar-Plumb CDN Providers requ
Re: (Score:2)
Netflix used to use Akamai, but then Level3 under-bid them. Akamai is a local cache CDN while Level3 is a Tier1 back-bone that uses it's massive cheap trunks to compete. It is cheaper for Level3 to deliver Netflix over the back-bone than for Akamai to cache the data locally. Akamai is still useful because not all Netflix customer's ISPs have peering with L3.
Because Level3 is acting as a CDN in the case of Netflix, this means all of the cost of Netfl
BBC has done this for iPlayer for a long time (Score:1)
Re: (Score:2)
Actually that changed in 2008, when the Beeb shifted larger content from the Akamai CDN to direct delivery over a peering agreement with Level3, causing much outrage amongst smaller ISPs.
http://www.theregister.co.uk/2008/08/22/bbc_cdn_isps_level3/ [theregister.co.uk]
The Good News... (Score:2)
...is that if content is mirrored inside your provider's network, it's frequently unmetered and thus doesn't count towards your monthly bandwidth limit...
Re: (Score:2)
Not. The hosted content is with BGP peering. Even though it is often on local 10gb links, it is still seen as a peer outside of your ISP. See NetFlix [netflix.com]
Edge? (Score:3)
The Internet has an edge? How can that be, the world is a sphere. (And theres no real final frontier of the internet in space yet,.
Re: (Score:3)
Re: (Score:3)
Bitter? (Score:4, Insightful)
Message from beyond our known civilization (Score:5, Insightful)
I would like to visit his planet. It sounds nice.
Re: (Score:2)
I would like to visit his planet. It sounds nice.
Content providers will have to subsidize, and fees/limits for accessing non-subsidizing content providers will be to the customer's disadvantage.
High speed subscription rates will be increased "Because see, we have better connectivity to Google, we're providing you a better service, you should pay more"
Quote from the article (Score:2, Flamebait)
This could help companies like Netflix get high definition video to consumers, and it could possibly bring down subscription rates for high speed internet, subsidized by the content providers.
BWHAHAHAHA!
Thanks for that, I needed a good laugh.
pipes pipes pipes (Score:2)
This isn't what this article is about, but this story reminds me that I would love to see some corporation, say Google, challenge the growing hegemony of telecoms and cable TV companies owning the Internet.
If we're going to go through all the trouble of building an Internet to replace TV and land line telephones (among other things) then it doesn't make any sense to hand the whole thing over to telecoms and cable companies.
Remember how all of a sudden all those small local ISPs just up and disappeared? We
Re: (Score:2)
It doesn't make sense to hand the whole thing to Google and a handful of other CDN companies.
dZ.
Re: (Score:2)
Remember how all of a sudden all those small local ISPs just up and disappeared? We once had tons of choices for Internet connectivity, now we've got a choice of maybe three corporations, all of whom hate us.
It's mostly about access to inexpensive pipes into subscribers homes. Before DSL/Cable technologies were developed, access was with dial-up modem, access to the subscribers was easy and inexpensive through POTS service, which is available at a flat rate, thanks to government regulations.
Cable and
Re: (Score:2)
ILECs have been disappearing at a pretty high rate. AT&T does not generally negotiate in good faith with them, and the rules for offering naked DSL are stringent. I think there about 1/2 as many ILECs as there were in 2006.
Re: (Score:2)
ILECs have been disappearing at a pretty high rate. AT&T does not generally negotiate in good faith with them
Huh? ATT is the ILEC at many exchanges.
Flying Monkeys (Score:2)
Bring Down Subscription Rates (Score:2)
..., and it could possibly bring down subscription rates for high speed internet, ...
Yeah, I don't see that statement as being true. Large web companies will only provide their on-site servers to large ISPs. The large ISPs have no reason to or history of reducing their subscription rates. If the servers were provided to smaller ISPs (such as GWI), then they could lower their rates and become more competitive. Maybe if the servers were provided to tier 2 peering networks, they could pass on the savings to small ISPs? I don't know enough about tiers or peering to know if that's a possibility
so regular CDNs are failing in the marketplace (Score:2)